1. Field of the Invention
The present invention relates to a laser apparatus, and more particularly it relates to an ultraviolet laser apparatus for generating ultraviolet light capable of suppressing generation of speckle with low coherence, such as an exposure light source used in a photo-lithography process for manufacturing micro devices such as semi-conductor elements, liquid crystal display elements, CCD and thin film magnetic heads, as well as relates to an exposure apparatus using such an ultraviolet laser apparatus.
2. Description of the Related Art
As information technology equipment has progressed, regarding integrated circuits, improvement in function, memory capacity and compactness have been requested, and, to achieve this, it is required that the degree of the integration of the integrated circuit be increased. In order to increase the degree of the integration, individual circuit patterns should be made smaller. However, a minimum pattern dimension of the circuit is generally determined by performance of an exposure apparatus used in a circuit manufacturing process.
In an exposure apparatus utilizing photo-lithography, a circuit pattern exactly described on a photo-mask is optically projected and transferred, with reduced scale, onto a semiconductor wafer on which photoresist is coated. A minimum pattern size (resolving power) R on the wafer in the exposure is represented by the following equation (1) and the depth of focus DF is represented by the following equation (2) when it is assumed that a wavelength of a light source used for projection in the exposure apparatus is λ and a numerical aperture of a projection optical system is NA:R=K·λ/NA  (1)DF=λ/{2·(NA)2}  (2)where, K is a constant.
As apparent from the above equation (1), in order to decrease the minimum pattern size R, the constant K may be decreased or the numerical aperture NA may be increased or the wavelength λ may be decreased.
Here, the constant K is a constant determined by the projection optical system or process and normally has a value of about 0.5 to 0.8. A method for decreasing the constant K is referred to as a super-resolution technique in a broader sense.
Regarding such a technique, an improvement in the projection optical system, modified illumination and a phase shift mask method have been proposed and investigated. However, they had disadvantage that applicable patterns were limited. On the other hand, from the above equation (1), the greater the numerical aperture NA the smaller the minimum pattern size R. However, this also means that the depth of focus is decreased, as apparent from the above equation (2). Thus, there is a limit to increase the numerical aperture NA, and, in consideration of the balance between NA and DF, the value of the numerical aperture NA is normally selected to about 0.5 to 0.6.
Accordingly, a most simple and effective method for decreasing the minimum pattern size is a method for decreasing the wavelength λ used in the exposure. There are several conditions in achieving reduction of the wavelength and in manufacturing the light source of the exposure apparatus. Now, these conditions will be described.
In a first condition, light output of several watts is required for shortening a time period for exposing and transferring the integrated circuit pattern.
In a second condition, in case of ultraviolet light having a wavelength smaller than 300 nm, material used for forming a lens of the exposure apparatus is limited, and it is difficult to correct chromatic aberration. Thus, monochromaticity of the light source is required and spectral width of must be smaller than 1 pm.
In a third condition, as spectral width is made narrower, temporal coherence is increased. Therefore, if light having a narrow line width is emitted as it is, an undesired interference pattern called as speckle will be generated. Accordingly, in order to suppress occurrence of the speckle, the spatial coherence in the light source must be reduced.
In order to satisfy these conditions and to realize high resolving power, many attempts for decreasing the wavelength of the exposure light source have been made. Heretofore, reduction of the wavelength has been investigated mainly in the following two ways. One way is a development to apply an excimer laser having a short oscillation wavelength to the exposure apparatus, and the other way is a development of a short wavelength exposure light source utilizing harmonic wave generation from an infrared or visual laser.
Among them, as the short wavelength light source realized by using the former way, a KrF excimer laser (wavelength of 248 nm) is known, and, nowadays, an exposure apparatus using an ArF excimer laser (wavelength of 193 nm) as a shorter wavelength light source is being developed. However, these excimer lasers hate several disadvantages that they are bulky, that optical parts are apt to be damaged because of great energy per one pulse and that maintenance of the laser is troublesome and expensive because of usage of harmful fluorogas.
On the other hand, as the latter way, there is a method for converting long wavelength light (infrared light or visual light) into shorter wavelength ultraviolet light by utilizing secondary non-linear optical effect of non-linear optical crystal. For example, in the document “Longitudinally diode pumped continuous wave 3.5 W green laser” (L. X, Liu, M. Oka, W. Wiechmann and S. Kubota, Optic Letters. vol. 19 (1994), p 189), a laser light source for wavelength-converting light from a solid-state laser of semiconductor excitation type is disclosed. In this conventional example, a laser beam having a wavelength of 1064 nm and emitted from an Nd:YAG laser is wavelength-converted by using the non-linear optical crystal to thereby generate 4th harmonic light having a wavelength of 266 nm. Further, the “solid-state laser” is a general term of lasers in which a laser medium is solid. Accordingly, although a semiconductor laser is included in the solid-state laser in a broad sense, normally, the solid-state laser means lasers excited by light such as a ND:YAG laser and a ruby laser, and, thus, in this specification, such a definition is used.
Further, as an example that the solid-state laser is used as the light source of the exposure apparatus, an array laser in which a plurality of laser elements each comprising a laser generating portion for generating a laser beam and a wavelength converting portion for wavelength-converting the light from the laser generating portion into ultraviolet light are bundled in a matrix patterns has been proposed. For example, Japanese Patent Laid-open No. 8-334803 (1996) discloses an example of an array laser in which a plurality of laser elements for wavelength-converting light from a laser generating portion having a semiconductor laser into ultraviolet light by using non-linear optical crystal provided in a wavelength converting portion are bundled in a matrix pattern (for example, 10×10) to thereby form a single ultraviolet light source.
According to the array laser having the above-mentioned arrangement, by bundling the plurality of independent laser elements together, light output of the entire apparatus can be increased while keeping light output of the individual laser element at a lower level. Thus, the load to the non-linear optical element can be reduced. However, since the laser elements are independent, when they are applied to the exposure apparatus, as a whole, oscillation spectra of the laser elements must be coincided For example, even when the line width of the oscillation spectrum of each laser elements is smaller than 1 pm, the difference in relative wavelength in the entire assembly including the plural laser elements must not be 3 pm, and the entire width must be smaller than 1 pm.
To achieve this, for example, lengths of resonators of the laser elements must be adjusted or wavelength selecting elements must be inserted into the resonators so that the laser elements each can independently perform single longitudinal mode oscillation having the same wavelength. However, these methods have disadvantages that the adjustment is delicate and that, as the number of laser elements is increased, the arrangement for causing all of the laser elements to perform oscillation having the same wavelength becomes more complicated.
On the other hand, as a method for actively equalizing the wavelengths from the plurality of laser elements, an injection seed method is well known (for example, refer to a document “Solid-state Laser Engineering”, 3rd Edition, Springer Series in Optical Science, Vol. 1, Springer-Verlag, ISBN 0-387-53756-2, p 246–249 presented by Walter Koechner). This method is a technique in which light from a single laser light source having narrow oscillation spectrum line width is branched to a plurality of laser elements and oscillation wavelengths of the laser elements coincide or are tuned by using the laser beams as seed light, thereby making the line widths of the spectra narrower. However, this method has a disadvantage that the arrangement becomes complicated, since an optical path for branching the seed light into the laser elements and a tuning and controlling portion for the oscillation wavelengths are required.
Further, although such an array laser can make the entire apparatus smaller considerably in comparison with the conventional excimer lasers, it is still difficult to obtain a packaging capable of suppressing output beam diameter of the entire array to less than several centimeters. Further, in the array laser having such an arrangement, there arise problems that the laser is expensive because the wavelength converting portions are required for the respective arrays and that, if mis-alignment occurs between the laser elements constituting the array or if the optical element(s) are damaged, in order to adjust the laser elements, the entire array must once be disassembled to remove the laser elements and the removed array must be assembled again after adjustment thereof.